MadLisp

MadLisp is a Lisp interpreter written in PHP. It is inspired by the Make a Lisp project, but does not follow that convention or syntax strictly. It provides a fun platform for learning functional programming.

Goals

• REPL environment where the user can interactively experiment with the language. Suitable for executing pieces of code one by one and examining the internal state of the system.
• Minimal safeguards or restrictions as to what can be done. Breaking things or using the language in unexpected ways should be part of the fun.
• Performance does not need to match commercial-grade languages, but needs to be fast enough for real-world programs and uses cases.
• Suitable to be used as a scripting language in Linux shell scripts and similar environments.
• Suitable to be used as an embedded scripting language inside another PHP application.
• Clear and intuitive error messages. This is important for pleasant user experience.
• Provide a library with commonly used features such as HTTP requests, JSON processing and SQL database support.
• Provide a clean interface for extending the language with your own functions defined in PHP.
• Provide a safe-mode where access to the file system and other external I/O is restricted.
• Provide a debug mode which shows what is happening inside the code evaluation.
• Loosely respect the Lisp legacy with things like naming conventions but do not be constrained by it.

Non-goals

• Ability to call arbitrary PHP functions directly. The language should have control over which PHP functions can be called and how.
• Namespaces or similar mechanisms. The global namespace is a feature, not a bug! Use a prefix for your function names if this becomes a problem.

Requirements

The project requires PHP 7.4 or newer.

The core project does not have any dependencies to external Composer libraries, but it does currently use Composer for the autoloader so you need to run composer install for that.

Usage

Use the run.php file to invoke the interpreter from the command line. You can start the Repl with the -r switch:

$ php run.php -r

You can also evaluate code directly with the -e switch:

$ php run.php -e "(+ 1 2 3)"
6

You can evaluate a file by giving it as argument:

$ php run.php file.mad

With no arguments the script will read input from stdin:

$ echo "(+ 1 2 3)" | php run.php
6

Init file

You can create an init file in your home directory with the name .madlisp_init. This file is automatically executed when the interpreter is started. It is useful for registering commonly used functions and performing other initialization.

Using from PHP

You can use the LispFactory class to create an instance of the interpreter if you wish to embed the MadLisp language in your PHP application and call it directly from your code.

Safe-mode

The language features a safe-mode that disables functions which allow external I/O. This allows a "sandbox" to be created where the evaluated scripts do not have access to the file system or similar resources. It is intended to be used when MadLisp is used as an embedded scripting language inside another application.

Types

Numbers

Numeric literals are interpreted as integer or floating point values. For example 1 or 1.0.

Strings

Strings are limited by double quotes, for example "this is a string".

Comments

Comments start with semicolon ; and end on a newline character.

Keywords

Special keywords are true, false and null which correspond to same PHP values.

Sequences

Lists are limited by parenthesis. When they are evaluated, the first item of a list is called as a function with the remaining items as arguments. They can be defined using the built-in list function:

> (list 1 2 3)
(1 2 3)

Vectors are defined using square brackets or the built-in vector function:

> [1 2 3]
[1 2 3]

> (vector 4 5 6)
[4 5 6]

Internally lists and vectors are just PHP arrays wrapped in a class, and the only difference between the two is how they are evaluated. Another reason for adding vectors is the familiarity of the square bracket syntax for PHP developers. They can be thought of as PHP arrays for most intents and purposes.

Hash maps

Hash maps are collections of key-value pairs. Keys are normal strings, not "keywords" starting with colon characters as in many Lisp languages.

Hash maps are defined using curly brackets or using the built-in hash function. Odd arguments are treated as keys and even arguments are treated as values. The key-value pair can optionally include colon as a separator to make it more readable, but it is ignored internally.

> (hash "a" 1 "b" 2)
{"a":1 "b":2}

> {"key":"value"}
{"key":"value"}

Internally hash maps are just regular associative PHP arrays wrapped in a class.

Symbols

Symbols are words which do not match any other type and are separated by whitespace. They can contain special characters. Examples of symbols are a, name or +.

Environments

Environments are hash-maps which store key-value pairs and use symbols as keys. Symbols are evaluated by looking up the corresponding value from the current environment. If the key is not defined in current environment the lookup proceeds to the parent environment and so forth. The initial environment is called root and contains all the built-in functions listed here. Then another environment called user is created for anything the user wants to define. The let and fn special forms create new local environments. Note that def always uses the current environment, so anything defined with def is not visible in the parent environment.

You can get the name of an environment using the meta function:

> (meta (env) "name")
"root/user"

You can also retrieve the parent environment:

> (meta (env) "parent")
{}

Quoting

Use the quote special form to skip evaluation:

> (quote (1 2 3))
(1 2 3)

Use the quasiquote special form when you need to turn on evaluation temporarily inside the quoted element. The special forms unquote and unquote-splice are available for that purpose:

> (def lst (quote (2 3)))
(2 3)

> (quasiquote (1 lst 4))
(1 lst 4)
> (quasiquote (1 (unquote lst) 4))
(1 (2 3) 4)
> (quasiquote (1 (unquote-splice lst) 4))
(1 2 3 4)

Internally quasiquote expands to cons and concat functions. We can use the quasiquote-expand special form to test this expansion without evaluation:

> (def lst (quote (2 3)))
(2 3)

> (quasiquote-expand (1 lst 4))
(cons 1 (cons (quote lst) (cons 4 ())))
> (quasiquote-expand (1 (unquote lst) 4))
(cons 1 (cons lst (cons 4 ())))
> (quasiquote-expand (1 (unquote-splice lst) 4))
(cons 1 (concat lst (cons 4 ())))

Quote shortcuts

You can use the single-quote ('), backtick and tilde (~) characters as shortcuts for quote, quasiquote and unquote respectively:

> '(a b c)
(a b c)

> `(a ~(+ 1 2) c)
(a 3 c)

All special forms related to quoting require exactly one argument.

Macros

The language has support for Lisp-style macros. Macros are like preprocessor directives and allow the manipulation of the language syntax before evaluation.

There are two built-in macros: defn which is a shortcut for the form (def ... (fn ...)) and defmacro which is a shortcut for the form (def ... (macro ...)).

We can use the special form macroexpand to test macro expansion without evaluating the resulting code. To illustrate how macros work, lets use defn as an example, and then view the expanded form using macroexpand:

> (def defn (macro (name args body) (quasiquote (def (unquote name) (fn (unquote args) (unquote body))))))
<macro>
> (macroexpand (defn add (a b) (+ a b)))
(def add (fn (a b) (+ a b)))

For another example, lets combine if and not into a macro named unless, this time using a shorter syntax:

> (defmacro unless (pred a b) `(if (not ~pred) ~a ~b))
<macro>
> (macroexpand (unless false "is false" "not false"))
(if (not false) "is false" "not false")
> (unless false "is false" "not false")
"is false"

The quasiquote form described above is essential for declaring macros. Internally macros are just functions with a special flag.

Exceptions

The language has support for try-catch style exception handlers. The syntax is (try A (catch B C)) where A is evaluated first and if exception is thrown, then C is evaluated with the symbol B bound to the value of the exception. Exceptions are thrown using the throw core function. You can give any data structure as argument to throw and it is passed along to catch. This way exceptions can contain more data than just a string that represents an error message.

Simple example of throwing and catching an exception:

> (try (throw {"msg":"message"}) (catch ex (str "error: " (get ex "msg"))))
"error: message"

Exceptions generated by PHP are catched as well. Their value will be a hash-map with keys type, file, line and message:

> (try (get "wrong" "type") (catch ex (get ex "type")))
"TypeError"

The Repl contains its own exception handler defined in PHP that will catch any exceptions thrown outside of try-catch form.

Reflection

You can use the meta special form to retrieve the arguments, body, code or full definition of user-defined functions:

> (defn add (a b) (+ a b))
<function>

> (meta add "args")
(a b)
> (meta add "body")
(+ a b)
> (meta add "code")
(fn (a b) (+ a b))
> (meta add "def")
(defn add (a b) (+ a b))

This allows for some fun tricks. For example, we can retrieve the body of a function and evaluate it as part of another function:

> (defn addOne (n) (+ n 1))
<function>
> (defn addTwo (n) (+ n 2))
<function>
> (defn addBoth (n) (+ (eval (meta addOne "body")) (eval (meta addTwo "body"))))
<function>
> (addBoth 1)
5

Special forms

Name	Safe-mode	Example	Example result	Description
and	yes	(and 1 0 2)	0	Return the first value that evaluates to false, or the last value.
case	yes	(case (= 1 0) 0 (= 1 1) 1)	1	Treat odd arguments as tests and even arguments as values. Evaluate and return the value after the first test that evaluates to true.
yes	(case (= 1 0) 0 "no match")	"no match"	You can also give optional last argument to case which is returned if none of the tests evaluated to true.
def	yes	(def addOne (fn (a) (+ a 1)))	<function>	Define a value in the current environment.
do	yes	(do (print 1) 2)	12	Evaluate multiple expressions and return the value of the last.
env	yes	(env +)	<function>	Return a definition from the current environment represented by argument. Without arguments return the current environment as a hash-map.
eval	yes	(eval (quote (+ 1 2)))	3	Evaluate the argument.
fn	yes	(fn (a b) (+ a b))	<function>	Create a function. Arguments can also be given as a vector instead of a list.
if	yes	(if (< 1 2) "yes" "no")	"yes"	If the first argument evaluates to true, evaluate and return the second argument, otherwise the third argument. If the third argument is omitted return null in its place.
let	yes	(let (a (+ 1 2)) a)	3	Create a new local environment using the first argument (list) to define values. Odd arguments are treated as keys and even arguments are treated as values. The last argument is the body of the let-expression which is evaluated using this new environment.
load	no	(load "file.mad")		Read and evaluate a file. The contents are implicitly wrapped in a do expression.
macro	yes			See the section Macros.
macroexpand	yes			See the section Macros.
meta	yes			See the sections Environments and Reflection.
or	yes	(or false 0 1)	1	Return the first value that evaluates to true, or the last value.
quote	yes			See the section Quoting.
quasiquote	yes			See the section Quoting.
quasiquote-expand	yes			See the section Quoting.
try	yes			See the section Exceptions.
undef	yes	(undef myFn)	<function>	Remove a definition from the current environment. Return the removed value.

Functions

Core functions

Name	Safe-mode	Example	Example result	Description
debug	no	(debug)	true	Toggle debug output.
doc	yes	(doc +)	"Return the sum of all arguments."	Show the documentation string for a function.
yes	(doc myfn "Documentation string.")	"Documentation string."	Set the documentation string for a function.
exit	no	(exit 1)		Terminate the script with given exit code using exit.
loop	yes	(loop (fn (a) (do (print a) (coinflip))) "hello ")	hello hello hello false	Call the given function repeatedly in a loop until it returns false.
print	no	(print "hello world")	hello world	Print expression on the screen. Give optional second argument as true to show strings in readable format. Print returns null (shown in REPL).
pstr	yes	(pstr {"a":"b"})	"{\"a\":\"b\"}"	Print expression to string.
read	yes	(read "(+ 1 2 3)")	(+ 1 2 3)	Read a string as code and return the expression.
sleep	no	(sleep 2000)	null	Sleep for the given period given in milliseconds using usleep.
throw	yes	(throw "invalid value")	error: "invalid value"	Throw an exception. The given value is passed to catch. See the section Exceptions.
timer	no	(timer (fn (d) (sleep d)) 200)	0.20010209	Measure the execution time of a function and return it in seconds.

Collection functions

Name	Example	Example result	Description
hash	(hash "a" 1 "b" 2)	{"a":1 "b":2}	Create a new hash-map.
list	(list 1 2 3)	(1 2 3)	Create a new list.
vector	(vector 1 2 3)	[1 2 3]	Create a new vector.
range	(range 2 5)	[2 3 4]	Create a vector with integer values from first to argument (inclusive) to second argument (exclusive).
range	(range 5)	[0 1 2 3 4]	Range can also be used with one argument in which case it is used as length for a vector of integers starting from 0.
ltov	(ltov '(1 2 3))	[1 2 3]	Convert list to vector.
vtol	(vtol [1 2 3])	(1 2 3)	Convert vector to list.
empty?	(empty? [])	true	Return true if collection is empty, otherwise false.
get	(get [1 2 3] 0)	1	Return the nth element from a sequence, or the corresponding value for the given key from a hash-map.
len	(len [1 2 3])	3	Return the number of elements in a collection.
first	(first [1 2 3 4])	1	Return the first element of a sequence.
second	(second [1 2 3 4])	2	Return the second element of a sequence.
penult	(penult [1 2 3 4])	3	Return the second-last element of a sequence.
last	(last [1 2 3 4])	4	Return the last element of a sequence.
head	(head [1 2 3 4])	[1 2 3]	Return new sequence which contains all elements except the last.
tail	(tail [1 2 3 4])	[2 3 4]	Return new sequence which contains all elements except the first.
slice	(slice [1 2 3 4 5] 1 3)	[2 3 4]	Return a slice of the sequence using offset and length. Uses array_slice.
apply	(apply + 1 2 [3 4])	10	Call the first argument using a sequence as argument list. Intervening arguments are prepended to the list.
chunk	(chunk [1 2 3 4 5] 2)	[[1 2] [3 4] [5]]	Divide a sequence to multiple sequences with specified length using array_chunk.
concat	(concat [1 2] '(3 4))	(1 2 3 4)	Concatenate multiple sequences together and return them as a list.
push	(push [1 2] 3 4)	[1 2 3 4]	Create new sequence by inserting arguments at the end.
cons	(cons 1 2 [3 4])	[1 2 3 4]	Create new sequence by inserting arguments at the beginning.
map	(map (fn (a) (* a 2)) [1 2 3])	[2 4 6]	Create new sequence by calling a function for each item. Uses array_map internally.
map2	(map2 + [1 2 3] [4 5 6])	[5 7 9]	Create new sequence by calling a function for each item from both sequences.
reduce	(reduce + [2 3 4] 1)	10	Reduce a sequence to a single value by calling a function sequentially of all arguments. Optional third argument is used to give the initial value for first iteration. Uses array_reduce internally.
filter	(filter odd? [1 2 3 4 5])	[1 3 5]	Create a new sequence by using the given function as a filter. Uses array_filter internally.
filterh	(filterh (fn (v k) (prefix? k "a")) {"aa":1 "ab":2 "bb":3})	{"aa":1 "ab":2}	Same as filter but for hash-maps. First argument passed to the callback is the value and second is the key.
reverse	(reverse [1 2 3])	[3 2 1]	Reverse the order of a sequence. Uses array_reverse internally.
key?	(key? {"a" "b"} "a")	true	Return true if the hash-map contains the given key.
set	(set {"a" 1} "b" 2)	{"a":1 "b":2}	Create new hash-map which contains the given key-value pair.
set!	(set! {"a" 1} "b" 2)	2	Modify the given hash-map by setting the given key-value pair and return the set value. This function is mutable!
unset	(unset {"a":1 "b":2 "c":3} "b")	{"a":1 "c":3}	Create a new hash-map with the given key removed.
unset!	(unset! {"a":1 "b":2 "c":3} "b")	2	Modify the given hash-map by removing the given key and return the corresponding value. This function is mutable!
keys	(keys {"a" 1 "b" 2})	("a" "b")	Return a list of the keys for a hash-map.
values	(values {"a" 1 "b" 2})	(1 2)	Return a list of the values for a hash-map.
zip	(zip ["a" "b"] [1 2])	{"a":1 "b":2}	Create a hash-map using the first sequence as keys and the second as values. Uses array_combine internally.
sort	(sort [6 4 8 1])	[1 4 6 8]	Sort the sequence using sort.
usort	(usort (fn (a b) (if (< a b) 0 1)) [3 1 5 4 2])	[1 2 3 4 5]	Sort the sequence using custom comparison function using usort.

Comparison functions

Name	Example	Example result	Description
=	(= 1 "1")	true	Compare arguments for equality using the == operator in PHP.
==	(== 1 "1")	false	Compare arguments for strict equality using the === operator in PHP.
!=	(!= 1 "1")	false	Compare arguments for not-equality using the != operator in PHP.
!==	(!== 1 "1")	true	Compare arguments for strict not-equality using the !== operator in PHP.
<	(< 1 2)	true	Return true if first argument is less than second.
<=	(<= 1 2)	true	Return true if first argument is less or equal to second.
>	(> 1 2)	false	Return true if first argument is greater than second.
>=	(>= 1 2)	false	Return true if first argument is greater or equal to second.

Database functions

This is a simple wrapper for PDO. This library is disabled in safe-mode.

Name	Example	Example result	Description
db-open	(def d (db-open "mysql:host=localhost;dbname=test" "testuser" "testpw"))	<object<PDO>>	Open a database connection.
db-execute	(db-execute d "INSERT INTO test_table (col1, col2) values (?, ?)" [1, 2])	1	Execute a SQL statement and return the number of affected rows.
db-query	(db-query d "SELECT * FROM test_table WHERE col1 = ?" [1])		Execute a SELECT statement.
db-last-id	(db-last-id d)	"1"	Return the last id of auto-increment column.
db-trans	(db-trans d)	true	Start a transaction.
db-commit	(db-commit d)	true	Commit a transaction.
db-rollback	(db-rollback d)	true	Roll back a transaction.

Http functions

This is a simple wrapper for cURL. This library is disabled in safe-mode.

Name	Example	Example result	Description
http	(http "POST" "http://example.com/" (to-json {"key":"value"}) {"Content-Type":"application/json"})	{"status":200 "body":"" "headers":{}}	Perform a HTTP request. First argument is the HTTP method, second is URL, third is request body and fourth is headers as a hash-map. The function returns a hash-map which contains keys status, body and headers.

IO functions

This library is disabled in safe-mode.

Name	Example	Example result	Description
wd	(wd)	"/home/pekka/code/madlisp/"	Get the current working directory.
chdir	(chdir "/tmp")	true	Change the current working directory.
file?	(file? "test.txt")	true	Return true if the file exists.
fget	(fget "test.txt")	"content"	Read the contents of a file using file_get_contents.
fput	(fput "test.txt" "content")	true	Write string to file using file_put_contents. Give optional third parameter as true to append.
fopen	(def f (fopen "test.txt" "w"))	<resource>	Open a file for reading or writing. Give the mode as second argument.
fclose	(fclose f)	true	Close a file resource.
fwrite	(fwrite f "abc")	3	Write to a file resource.
fflush	(fflush f)	true	Persist buffered writes to disk for a file resource.
fread	(fread f 16)	"abc"	Read from a file resource.
feof?	(feof? f)	true	Return true if end of file has been reached for a file resource.
readline	(readline "What is your name? ")	What is your name?	Read line of user input using readline.
readline-add	(readline-add "What is your name? ")	true	Add line of user input to readline history using readline_add_history.
readline-load	(readline-load "historyfile")	true	Read readline history from file using readline_read_history.
readline-save	(readline-save "historyfile")	true	Write readline history into file using readline_write_history.

Json functions

Name	Example	Example result	Description
to-json	(to-json { "a" [1 2 3] "b" [4 5 6] })	"{\"a\":[1,2,3],\"b\":[4,5,6]}"	Encode the argument as a JSON string.
from-json	(from-json "{\"a\":[1,2,3],\"b\":[4,5,6]}")	{"a":[1 2 3] "b":[4 5 6]}	Decode the JSON string given as argument.

Math functions

Name	Example	Example result	Description
+	(+ 1 2 3)	6	Return the sum of the arguments.
-	(- 4 2 1)	1	Subtract the other arguments from the first.
*	(* 2 3 4)	24	Multiply the arguments.
/	(/ 7 2)	3.5	Divide the arguments.
//	(// 7 2)	3	Divide the arguments using integer division.
%	(% 6 4)	2	Calculate the modulo.
inc	(inc 1)	2	Increment the argument by one.
dec	(dec 2)	1	Decrement the argument by one.
sin	(sin 1)	0.84	Calculate the sine.
cos	(cos 1)	0.54	Calculate the cosine.
tan	(tan 1)	1.55	Calculate the tangent.
abs	(abs -2)	2	Get the absolute value.
floor	(floor 2.5)	2	Get the next lowest integer.
ceil	(ceil 2.5)	3	Get the next highest integer.
pow	(pow 2 4)	16	Raise the first argument to the power of the second argument.
sqrt	(sqrt 2)	1.41	Calculate the square root.
coinflip	(coinflip)	true	Return true or false with equal probability.
rand	(rand 5 10)	8	Return a random integer between given min and max values.

Regular expression functions

Name	Example	Example result	Description
re-match	(re-match "/^[a-z]{4}[0-9]{4}$/" "test1234")	true	Match subject to regular expression using preg_match.
(re-match "/[a-z]{5}/" "one three five" true)	"three"	Give true as third argument to return the matched text.
re-match-all	(re-match-all "/[A-Z][a-z]{2}[0-9]/" "One1 Two2 Three3")	["One1" "Two2"]	Find multiple matches to regular expression using preg_match_all.
re-replace	(re-replace "/year ([0-9]{4}) month ([0-9]{2})/" "$1-$2-01" "year 2020 month 10")	"2020-10-01"	Perform search and replace with regular expression using preg_replace.
re-split	(re-split "/\\s+/" "aa bb cc ")	["aa" "bb" "cc"]	Split the subject by regular expression using preg_split. The flag PREG_SPLIT_NO_EMPTY is enabled.

String functions

Name	Example	Example result	Description
empty?	(empty? "")	true	Return true if argument is empty string.
len	(len "hello world")	11	Return the length of a string using strlen.
trim	(trim " abc ")	"abc"	Trim the string using trim.
upcase	(upcase "abc")	"ABC"	Make the string upper case using strtoupper.
lowcase	(lowcase "Abc")	"abc"	Make the string lower case using strtolower.
substr	(substr "hello world" 3 5)	"lo wo"	Get a substring using substr.
replace	(replace "hello world" "hello" "bye")	"bye world"	Replace substrings using str_replace.
split	(split "-" "a-b-c")	["a" "b" "c"]	Split string using explode.
join	(join "-" "a" "b" "c")	"a-b-c"	Join string together using implode.
format	(format "%d %.2f" 56 4.5)	"56 4.50"	Format string using sprintf.
prefix?	(prefix? "hello world" "hello")	true	Return true if the first argument starts with the second argument.
suffix?	(suffix? "hello world" "world")	true	Return true if the first argument ends with the second argument.

Note that support for multibyte characters in strings is limited because the provided functions do not use the mbstring extension.

Time functions

Name	Example	Example result	Description
time	(time)	1592011969	Return the current unix timestamp using time.
date	(date "Y-m-d H:i:s")	"2020-06-13 08:33:29"	Format the current time and date using date.
strtotime	(strtotime "2020-06-13 08:34:47")	1592012087	Parse datetime string into unix timestamp using strtotime.

Type functions

Skipped examples here as these are pretty self-explanatory.

Name	Description
bool	Convert the argument to boolean.
float	Convert the argument to floating-point value.
int	Convert the argument to integer.
str	Convert the argument to string. Also concatenate multiple strings together.
symbol	Convert the argument to symbol.
not	Turns true to false and vice versa.
type	Return the type of the argument as a string.
fn?	Return true if the argument is a function.
macro?	Return true if the argument is a macro.
list?	Return true if the argument is a list.
vector?	Return true if the argument is a vector.
seq?	Return true if the argument is a sequence (list or vector).
hash?	Return true if the argument is a hash-map.
symbol?	Return true if the argument is a symbol.
object?	Return true if the argument is an object.
resource?	Return true if the argument is a resource.
bool?	Return true if the argument is a boolean value (strict comparison).
true?	Return true if the argument evaluates to true (non-strict comparison).
false?	Return true if the argument evaluates to false (non-strict comparison).
null?	Return true if the argument is null (strict comparison).
int?	Return true if the argument is an integer.
float?	Return true if the argument is a floating-point value.
str?	Return true if the argument is a string.
zero?	Return true if the argument is integer 0 (strict comparison).
one?	Return true if the argument is integer 1 (strict comparison).
even?	Return true if the argument is even number (0, 2, 4, ...).
odd?	Return true if the argument is odd number (1, 3, 5, ...).

Constants

The following constants are defined by default:

Name	Value
DIRSEP	PHP constant DIRECTORY_SEPARATOR
HOME	PHP constant $_SERVER['HOME']
EOL	PHP constant PHP_EOL
PI	PHP constant M_PI
__DIR__	Directory of a file being evaluated using the special form load. Otherwise null.
__FILE__	Filename of a file being evaluated using the special form load. Otherwise null.

Extending

The project is easy to extend because it is trivial to add new functions whether the implementation is defined on the PHP or Lisp side.

License

MIT